Owing to the dynamic and competitive business environment, manufacturing companies take real-time monitoring and rapid decision making based on RFID applications, which brings huge volume of information and events generated in the defined manufacturing workflows. Complex Event Processing (CEP) is introduced to solve the problems mentioned above. CEP is applied to handle diverse and large amount of low-level multiple data and primitive events for the purpose of identifying meaningful event patterns. It is very important to integrate the CEP technology to the manufacturing workflows. In this paper, we provide a novel framework of RFID-based complex event processing system for assembly manufacturing applications like cars and high-speed trains. It bridges the hardware in workshops and enterprise applications. The Complex Event Management System (CEMS), which is the kernel of the framework, can filter the irrelevant events and work with uncertain data. Furthermore, a concrete example is used to describe the framework and validate the feasibility in assembly monitoring of the car manufacturing.